Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
M4+ Chip Generation - Speculation Megathread [MERGED]
- Thread starter GoetzPhil
- Start date
- Sort by reaction score
How about having an UltraFusion "die" that actually carries the now extra hardware? So instead of just having A+A, it is now an A+B+A.
That base M4 Max config… best value, next to base Mini. Quite tempting. I hope I can hold off until M5 Max though. Will have to keep that computer for at least 5 years.
No SME on M3 Ultra is gonna be…interesting.
Is it possible that Apple has changed the node and is using TSMC's N3E?
That’s where the remaining known Sequoia identifiers come into play:
Mac16,9 = M4 Max Mac Studio
Mac15,2 = M3 Ultra Mac Studio (on the theory that Mac15,1 was an M3 Max Mac Studio prototype)
Mac17,1 = M5 Ultra Mac Pro
Mac17,2 = M5 Ultra+ Mac Pro
I had heard Mac15,14 for the M3 Ultra but who knows at this point.
It has long been reported that designs on N3B are not compatible to fab on N3E.
If that were the case, why not just go with the M4 Ultra instead?
I think why not just go with the M4 Ultra is a question everyone is asking anyway! 🙃 But yeah I don't think a node shift is likely - as @Chancha said, reportedly N3B and N3E are not design compatible so that would seemingly rule out a node shift. Tweaking the die for TB5 and ultra fusion is one thing, completely redesigning it another.
Mac Pro is going to be eye watering expensive if it has a new chip. Whew what an interesting decision this was, real glad I went with a MBP and didn’t wait.
Looking forward to WWDC / fall announcements.
At least a new Studio Display / XDR must be coming since they’ve gone out of their way to make sure every high-end pro product has TB5.
Looking forward to WWDC / fall announcements.
At least a new Studio Display / XDR must be coming since they’ve gone out of their way to make sure every high-end pro product has TB5.
If M3 Ultra is 1.5 cpu performance of M2 Ultra, that’s about 30000 GB multi core score. M4 Max is 27000. Not that great if true.
A puzzling release. Huge LLMs seem to be the main beneficiaries.
A puzzling release. Huge LLMs seem to be the main beneficiaries.
It's possible it might have 1.5 the performance of the M4 Max depending on the benchmark - remember for more workstation type tasks, embarrassingly parallel tasks begin to dominate again so GB 6's desktop-oriented-take-the-average-of-embarrassingly-parallel-and-work-sharing-task approach may not be as applicable here. For instance, the M4 Max gets just over 2000pts in CB R24 and the the M3 Max was just over 1600pts. The full M3 Ultra could easily get 3000pts. I would imagine certain subscores of GB6, like compilation, will behave similarly. Also to be fair, the M3 Ultra may not have SME but it still has the AMX cores so for those processes which can use the Apple Accelerate framework, they should still see some benefit.
Still, this is all weird. If they were gonna release M3 Ultra, which they evidently have done, why didn't it come sooner? Why didn't the Studio and the Pro get M3 Max and Ultra both a year ago?
All I can think of is that Apple hesitated about the future state of the Mac Pro. Probably didn't sell well next to the Studio and therefore updating it to M3 Max & Ultra would be waste. At the same time, updating Studio to M3 and not the Mac Pro would definitely put a nail in the coffin for it.
So they, maybe just 6-12 months ago, decided to differentiate it further. Studio get's on a late cycle and Mac Pro a very early one. Mac Pro gets M5 series this WWDC. Studio will skip M5 series altogether if Mac Pro sells OK. If Mac Pro still sells bad, they will discontinue it, and then Studio could get the M5 Max and M5 Ultra as well in 2026.
All I can think of is that Apple hesitated about the future state of the Mac Pro. Probably didn't sell well next to the Studio and therefore updating it to M3 Max & Ultra would be waste. At the same time, updating Studio to M3 and not the Mac Pro would definitely put a nail in the coffin for it.
So they, maybe just 6-12 months ago, decided to differentiate it further. Studio get's on a late cycle and Mac Pro a very early one. Mac Pro gets M5 series this WWDC. Studio will skip M5 series altogether if Mac Pro sells OK. If Mac Pro still sells bad, they will discontinue it, and then Studio could get the M5 Max and M5 Ultra as well in 2026.
If a 40 Core/64GB Macbook Pro M4 Max is 10% Cheaper than the bottom spec 96GB/60 Core Mac Studio M3 Ultra then come October would you rather have:
+) 60 Core/96GB Mac Studio M3 Ultra or a
+) 40 Core/64GB Macbook Pro M5 Max
M4 GPU Core = (aprox) 25% faster than M3 GPU core
M5 GPU : 1.05 (N3E->N3P) x 1.05 (Guess architectural improvement) = 10%
M3 -> M5 : 1.25 x 1.10 = 38% Faster GPU core.
40 x M5 MAX GPU cores = Equiv to 40 x 1.38 = 55 x M3 ULTRA GPU cores.
+) And so come October the M5 Max Laptop will match the 60C M3 Ultra Studio in performance at the same price point.
As the Mac Pro will have come out at that point .. there is NO way Apple could charge 7k-8k ($ or £) ie roughly twice as the M5 Max Macbook Pro for the SAME performance.
And so perhaps will actually see a better value proposition for the entry level Mac Pro - because I doubt apple will offer the Mac Pro with M3 or M4 inside .. and they need to offer it with something Faster than M5 MAX to ensure than the Pro remains competitive with the M5 MAX laptops throughout 2026.
TLDR : My guess is that the LOWEST spec Mac Pro will be an M5 Ultra for £8k and with a pre-order in September 2025 for delivery in October.
+) 60 Core/96GB Mac Studio M3 Ultra or a
+) 40 Core/64GB Macbook Pro M5 Max
M4 GPU Core = (aprox) 25% faster than M3 GPU core
M5 GPU : 1.05 (N3E->N3P) x 1.05 (Guess architectural improvement) = 10%
M3 -> M5 : 1.25 x 1.10 = 38% Faster GPU core.
40 x M5 MAX GPU cores = Equiv to 40 x 1.38 = 55 x M3 ULTRA GPU cores.
+) And so come October the M5 Max Laptop will match the 60C M3 Ultra Studio in performance at the same price point.
As the Mac Pro will have come out at that point .. there is NO way Apple could charge 7k-8k ($ or £) ie roughly twice as the M5 Max Macbook Pro for the SAME performance.
And so perhaps will actually see a better value proposition for the entry level Mac Pro - because I doubt apple will offer the Mac Pro with M3 or M4 inside .. and they need to offer it with something Faster than M5 MAX to ensure than the Pro remains competitive with the M5 MAX laptops throughout 2026.
TLDR : My guess is that the LOWEST spec Mac Pro will be an M5 Ultra for £8k and with a pre-order in September 2025 for delivery in October.
I think it will get very interesting with the M5. We clearly haven’t seen what the ”Hidra” fuss is about yet. What if they pull off ultrafusion x4, but the top model is still the Ultra, it just uses 4 regular M5 chips?
M5, 6P cores and 2E cores. 15 GPU cores.
Pro gets double, Max x3, Ultra x4
Speculating is too much fun. Exciting WWDC ahead.
M5, 6P cores and 2E cores. 15 GPU cores.
Pro gets double, Max x3, Ultra x4
Speculating is too much fun. Exciting WWDC ahead.
On the other hand, the M1 Ultra took 16 months to come out after the M1. The M2 Ultra took 12 after the M2. Using that yardstick, suddenly the M3 Ultra again taking 16 doesn't look so bad ... except obviously these days the Max launches with the base chip (unlike previously) and also it appear that Apple is moving to yearly or near yearly Mac-chip launches such that the M3 Ultra is possibly launching closer to the M5 than the M3. I also agree about the possibility of the Mac Pro getting an early M5. So yeah still very weird, but also weirdly not? 🙃
Judging from what Nvidia has done with its Grace Blackwell Superchips, x4 is perhaps more likely to be two M5 Ultras on a substrate (CoWoS-L). Assuming Apple sticks with the Max as the building block, hardly a safe assumption now.
One possible explanation for the M3 Ultra delay might be that the entire production, until recently, has gone into Private Cloud Compute hardware.
Going by your supposition the GPU core count in an Ultra drops from 80-core to 60-core...?